Image Classification With Pre-Trained Keras Model

This Python script is designed for image classification using a pre-trained deep learning model. It reads images from an input folder, predicts their class labels, and organizes them into class-specific folders in an output directory.

Features

• Loads a pre-trained Keras model and class labels.
• Processes images (resizing and normalization).
• Predicts the class of each image.
• Moves images to class-specific folders based on predictions.
• Supports a confidence threshold to handle uncertain predictions.

Prerequisites

• Python 3.x
• Required Python packages: Keras, Pillow (PIL), NumPy or Miniconda to manage them at once

Usage

First-time setup

Create the folder containing input images

mkdir -p images/input

Then put some images into that folder.

Customize the following constants in the main.py script

• MODEL_PATH: Path to the pre-trained Keras model file.
• LABEL_PATH: Path to the file containing class labels.
• INPUT_IMAGE_FOLDER: Path to the folder containing input images.
• OUTPUT_IMAGE_FOLDER: Path to the folder where organized images will be saved.
• NOT_SURE_CLASS: Default class name for images with low confidence.
• NOT_SURE_THRESHOLD: Confidence threshold for classifying as "Not Sure."

Install the required Python packages

pip install keras tensorflow Pillow

Or, pretty much convenient way with Miniconda:

conda env create -f environment.yml

Start the script

# If you use Miniconda, activate the environment first.
conda activate keras-tf-pillow

python main.py